In the field of semiconductor processes, the design rule is going into an age of 100 nm and the production form is on a transition from a mass production with a few models representative of a DRAM into a small-lot production with a variety of models such as a SOC (Silicon on chip). This results in an increase of a number, of processes, and thus an improvement in yield for each process becomes essential, which makes more important an inspection for a defect possibly occurring in each process. The present invention relates to an apparatus to be used in the inspection of a wafer after respective processes in the semiconductor processes, and in particular to an inspection method and apparatus using an electron beam and also to a device manufacturing method using the same.
In conjunction with a high integration of semiconductor device and a micro-fabrication of pattern thereof, an inspection apparatus with higher resolution and throughput has been desired. In order to inspect a wafer substrate of 100 nm design rule for any defects, a resolution in size equal to or finer than 100 nm is required, and the increased number of processes resulting from a high integration of the device has called for an increase in the amount of inspection, which consequently requires higher throughput. In addition, as a multi-layer fabrication of the device has been progressed, the apparatus has been further required to have a function for detecting a contact failure in a via for interconnecting wiring between layers (i.e., an electrical defect). In the current trend, an inspection apparatus of optical method has been typically used, but it is expected that an inspection apparatus using an electron beam may soon be of mainstream, substituting for the inspection apparatus of optical method in the viewpoint of resolution and of inspection for contact failure. The inspection apparatus of electron beam method, however, has a weak point in that the inspection apparatus of electron beam method is inferior to the inspection apparatus of optical method in view of throughput.
Accordingly, it would be desirable to develop an apparatus having higher resolution and throughput and being capable of detecting the electrical defects. It is known that the resolution in the inspection apparatus of optical method is limited to ½ of the wavelength of the light to be used, and it is about 0.2 μm for an exemplary case of a visible light having put to practical use.
On the other hand, in the method using an electron beam, typically a scanning electron microscope (SEM method) has been put to practice, wherein the resolution thereof is 0.1 μm and the inspection time is 8 hours per wafer (20 cm wafer). The electron microscope method further has a distinctive feature that it is able to inspect for any electrical defects (breaking of wire in the wirings, bad continuity, bad continuity of via). However, the inspection speed (sometime also referred to as inspection speed) thereof is very low, and so the development of an inspection apparatus with higher inspection speed has been expected.
Generally, since an inspection apparatus is expensive and a throughput thereof is rather lower as compared to other processing apparatuses, therefore the inspection apparatus has been used after an important process, for example, after the process of etching, membrane deposition, CMP (Chemical-mechanical polishing) planarization or the like.
The inspection apparatus of scanning method using an electron beam (SEM) will now be described. In the inspection apparatus of SEM method, the electron beam is contracted to be narrower (the diameter of this beam corresponds to the resolution thereof) and this narrowed beam is used to scan a sample so as to radiate it linearly. On the other hand, moving a stage in the direction normal to the scanning direction allows an observation region to be irradiated by the electron beam as a plane area. The scanning width of the electron beam is typically some 100 μm. Secondary electrons emitted from the sample by the irradiation of said contracted and narrowed electron beam (referred to as a primary electron beam) are detected by a detector (a scintillator plus photo-multiplier (i.e., photoelectron multiplier tube) or a detector of semiconductor type (i.e., a PIN diode type) or the like). A coordinate for an irradiated location and an amount of the secondary electrons (signal intensity) are combined and formed into an image, which is stored in a storage or displayed on a CRT (a cathode ray tube). The above description shows the principle of the SEM (scanning electron microscope), and defects in a semiconductor wafer (typically made of Si) in the course of process may be detected from the image obtained in this method. The inspection speed (corresponding to the throughput) is varied in dependence on an amount of primary electron beam (the current value), a beam diameter, and a response time of the detector. The beam diameter of 0.1 μm (which may be considered to be equivalent to the resolution), the current value of 100 nA, and the response time of the detector of 100 MHz are the currently highest values, and in the case using those values the inspection speed has been evaluated to be about 8 hours for one wafer having the diameter of 20 cm.
In order to improve the inspection apparatus of said SEM method to work at much higher speed (to increase the throughput), a new method referred to as an image projecting method has been suggested. According to this method, an observation region on a sample is irradiated in block by a primary electron beam (i.e., no scanning but an irradiation covering a certain area), and secondary electrons emitted from the irradiated region are formed into an image in block by a lens system on a detector (a micro-channel plate (MCP) plus fluorescent screen) as an image of electron beam. That image is used in a two-dimensional CCD (charge coupled device) or a TDI-CCD (Time Delayed Integration-CCD) to convert the image data into an electric signal, and from this image data, defects in the sample wafer (the semiconductor (Si) wafer in the course of process) may be detected.
Accordingly, there has arisen a demand for constructing an overall system for inspecting such as a substrate with high level of accuracy and efficiency by using a defect inspection apparatus of said image projecting method having an advantage of higher throughput. Almost no study has made clear an overall structure of such an inspection apparatus that, for example, feeds a sample to inspection as in a clean state to an irradiating location in the image projecting optical apparatus, allowing for an association with the other subsystems to be brought in alignment with it. In addition, in such an environment where the diameter of a wafer subject to inspection has been made larger and larger, there has arisen a demand that the subsystems also should be correspondingly modified to be suitable for the wafer having a large diameter.
In the inspection system, maintaining a vacuum atmosphere within the chamber is one of the important terms. In an apparatus such as a defect inspection apparatus that provides an ultraprecision processing, a stage for accurately positioning a sample in the vacuum atmosphere has been used, wherein in the case where said stage is required to be positioned highly accurately, one structure has been conventionally employed, in which the stage is supported in non-contact manner by a hydrostatic bearing. In this case, the vacuum level in a vacuum chamber is maintained by forming in an extent of the hydrostatic bearing a differential pumping mechanism for exhausting a high pressure gas so that the high pressure gas supplied from the hydrostatic bearing may not be directly exhausted into the vacuum chamber. Aiming at such a stage, specifically a stage including a combination of the hydrostatic bearing and the differential pumping mechanism has been proposed, as shown in FIGS. 29[A] and [B]. In this configuration, when the stage moves, guide planes 6a-7 and 7a-7 facing to the hydrostatic bearing 9-7 move forth and back between the high pressure gas atmosphere in the hydrostatic section and the vacuum environment within the chamber. For this period, the gas is adsorbed onto the guide plane during it being exposed to the high pressure gas atmosphere and the adsorbed gas is discharged once the guide plane is exposed to the vacuum environment, which will repeatedly occur. Owing to this, every time when the stage moves, there occurs such an event that the vacuum level within the chamber C is deteriorated, which disadvantageously has inhibited the processing including the aforementioned exposure, inspection and process by using the charged particle beam from being performed stably, or otherwise the sample has been contaminated.
Further, there have been such problems in the above-described stage including a combination of the hydrostatic bearing and the differential pumping mechanism that because of the differential pumping mechanism having been added, the structure has become more complicated and the reliability as a stage has decreased in contrast with the increased cost.
As for the electron beam apparatus of the image projecting method by itself, since a plurality of signals from a plurality of pixels on the sample surface can be captured all at once, therefore this method is advantageous in the point of the pattern inspection with the high throughput, while the method is problematic in the point that the sample may be charged due to a plurality of pixels being exposed to the irradiation of the electron beam all at once. On the other hand, in the case where a mark for positioning on the wafer is to be detected during the processes, a field of view may not necessarily be such wide that would be required by the image projection in the pattern inspection but a narrower field of view may be sufficient, wherein it is rather problematic that an insufficiently small pixel size may result in an insufficient mark detection accuracy.
Besides, for the MCP, as a total output charge amount (screen current×time period) is increased over a long-time use, the MCP multiplication factor is decreased, and therefore there has been a problem in that a defective image contrast may change or deteriorate with the same MCP applied voltage upon picking up the defective images successively for a long period in the defect inspection apparatus.
Further, an amplification factor in the image beam current amount is determined by a voltage applied between a first MCP and a second MCP and for example, the amplification factor should be 1×104 with the applied voltage of 1.4 kV. Additionally, a voltage on the order of 3 kV is applied between the second MCP and the fluorescent screen in order to inhibit the expansion of the image beam output from the second MCP. A detector of a conventional electron beam apparatus, in which a camera sensor and a vacuum flange have been separately formed, is disadvantageous in that it has a longer signal line, it is susceptive to signal latency or disturbance, and it prohibits the fast driving of the detector, which has been factors to decrease the throughput in the inspection (a process volume per time).
Further, to perform the defect inspection by using the electron beam, an emission current flow to an electron gun is required to be controlled so as to keep the contrast in the picked-up image at a constant level, and typically the emission current has been controlled by adjusting a voltage applied to the Wehnelt electrode made of such material as LaB6 (lanthanum hexaboride) disposed downstream to the electron gun. FIG. 15 is a graph illustrating a relationship between the voltage (in volt) applied to the Wehnelt electrode and the emission current (in microampere) of the electron gun, and it is seen from this graph that if the voltage applied to the Wehnelt electrode exceeds the level of −300 volt, the emission current increases rapidly.
However, if the electron gun operates for a long time under the condition that the applied voltage to the Wehnelt electrode is maintained at a constant level, an oxide film including La and B emitted from the electron gun may adhere to the inside of the Wehnelt electrode and form an insulating film thereon, which will be positively charged. This is because the electron emitted from the electron gun has an accelerating energy as much as the applied voltage to the Wehnelt electrode, and such electrons impinging upon said insulating film may cause the insulating film to emit the secondary electrons more than the electrons flowing into the Wehnelt electrode. As a matter of fact, there has been a problem in that the applied voltage to the Wehnelt electrode shifting to the positive direction causes a gradual increase in the emission current of the electron gun, which makes it difficult to hold the constant emission current.
On the other hand, advantageously the inspection apparatus having a function as the scanning electron microscope according to the prior art, as compared to the inspection apparatus of the image projecting method, has no such problem that the sample is charged but has a sufficient mark detection accuracy. Individually, either of them has to solve the following problems.
For example, if a sample wafer includes a via formed therein, then a care must be taken upon performing an evaluation procedure for the sample wafer. This is because if a large decelerating electric field as well as the primary electron beam by the amount of not less than a certain value is applied between the objective lens and the wafer, a discharge occurs between the via and the objective lens, and said discharge may possibly cause a damage to a device pattern formed in the wafer. There are a wafer of one type that is apt to incur such a discharge and a wafer of other type that hardly incurs the discharge, wherein the wafer of either type has a different condition of inducing the discharge (different decelerating electric field voltage value and different primary electron beam amount).
Further, it has been known that an edge portion of a pattern is apt to dazzle due to a higher secondary electron emission rate. With the higher secondary electron emission rate, the detection signal of the secondary electron beam to be output by the detector has an increasing signal intensity, and disadvantageously this detection signal results in a masking of a signal generated by a defect, thereby deteriorating the inspection speed.
In an apparatus for evaluating the post-process condition of a wafer according to the prior art, the inspection is performed to encompass the entire area of the wafer, and therefore the wafer is moved within a working chamber so that an arbitrary point on the wafer surface may be positioned in alignment with the optical axis of the electron beam. Accordingly, the evaluation apparatus of the conventional example needs a bottom area extended in the forth and back and the left and right directions by such an amount that can accommodate the movement of the wafer, and inevitably the evaluation apparatus has an enlarged floor area. This enlargement of the floor area is a counter trend toward the effective use of the clean room, thus it is desired to make the evaluation apparatus compact.
Further, said conventional apparatus requires an inspection time of a few hours for a single wafer (a few hours/wafer) to accomplish the inspection covering the entire surface of the wafer. On the other hand, the throughput in the wafer processing apparatus reaches to approximately 100 wafers in 1 hour (about 100 wafers/hour), which means that the inspection time of the wafer is equivalent to dozen times of the process time. Thus, in the circumstance that there is a mismatch between the throughput of the evaluation apparatus and the throughput of the processing apparatus, it would be desirable that those throughputs are matched to each other by reducing the inspection time.
A gate oxide of the semiconductor device is apt to be made thinner year by year, and it has been believed that the thickness of the film may be on the order of 1 nm in the year of 2005 and 0.5 nm in the year of 2005. In addition, the minimum line width “d” of the pattern formed in the sample subject to the inspection is getting narrower, and in proportion to that, it is required to reduce the pixel size used in the evaluation. On the other hand, in order to secure an S/N ratio of the signal at a certain level, it is required to obtain a certain amount of detected secondary electrons per pixel, consequently leading to the trend of increasing the amount of the primary electron per unit area. As a result, the gate oxide is likely to be damaged (including breakdown) as it is getting thinner, while the voltage generated between both sides of the gate oxide increases as the dose increases, thereby the gate oxide is more apt to be damaged. In this viewpoint, it is strongly desired that such an electron beam apparatus be provided which would not give any damage to a thin gate oxide of a sample to be inspected.
Besides, there is a need for improving the throughput as much as possible, and thus it is desired that a sample such as a single wafer (hereafter referred to as a sample for simplicity) may be completely inspected or evaluated within a process time as long as that taken by a process prior to the inspection process. In this regard, it is also conceivable that the inspection time per sample may be reduced by evaluating an arbitrary small number of chips among many chips in a single sample.
Further, there has been no study on how to inhibit an aberration due to forming the primary beam into a multi-beam. In specific, such a method is strongly required that forms the multi-beam by using an optical system which prevents an image field curvature aberration, which is a most serious aberration among aberrations associated with the primary beam.
In conjunction with a defect inspection apparatus described above which uses the image projecting system and the multi-beam scanning method, it has been also suggested that an image recognition technology should be improved so as to achieve a fast and highly accurate defect inspection on a patter of micro-fabrication. However, the prior art has a problem that there may be a position mismatch between an image of the secondary electron beam obtained by irradiating the primary electron beam onto a region to be inspected on a surface of a sample and a reference image which has been prepared in advance, thereby resulting in the deterioration in the accuracy in the defect inspection. This position mismatch could be a serious problem specifically when the irradiating region of the primary electron beam is in the misalignment with respect to the wafer and a part of the inspected pattern falls out of the detected image of the secondary electron beam, which could not be compensated for properly by simply using the technology for optimizing the matching region within the detected image. This problem could be a critical drawback specifically in the inspection of a pattern of high-precision.
An object of the present invention is to provide an inspection method and an apparatus using an electron beam which have overcome those aforementioned problems and can detect a defect in a sample with high level of throughput and accuracy, and also to provide a semiconductor device manufacturing method using these inspection method and apparatus.